Correcting [18F]MK‐6240 SUVR for non‐equilibrium effects using tissue clearance correction method does not improve its performance

TL;DR

A method to correct PET scan measurements for brain imaging did not improve results for a specific radiotracer.

Contribution

The study shows that a previously proposed correction method does not reduce bias variability for [18F]MK-6240 radiotracer.

Findings

SUVR correction did not significantly reduce variability in bias compared to uncorrected SUVR.

The method performed poorly for [18F]MK-6240, which does not follow a 1-tissue compartment model.

Original SUVR measurements were already minimally biased in this cohort.

Abstract

Honhar et al. (PMCID:10993874) reported a method for bias and noise correction in standardized‐uptake value ratios (SUVR, simplified outcome), relative to distribution volume ratios (DVR, quantitative gold‐standard), without the need for dynamic data outside SUVR time‐window. The method performed well for rapidly‐reversible radiotracers, [18F]FE‐PE2I and [11C]LSN3172176 (i.e., radiotracers kinetics described by 1‐tissue compartment or simplified reference tissue models). Herein, we tested this method on human [18F]MK‐6240 PET data.

Dynamic [18F]MK‐6240 PET data (0‐120 min) in 22 human participants (age: 61±18 y, 8 females, 16 cognitively unimpaired controls [HC], 5 mild cognitive impairment [MCI], 1 Alzheimer's disease [AD]) were reanalyzed. Metabolite‐corrected arterial input functions were used to compute regional 2‐tissue compartment model DVR values (reference: eroded cerebellar gray matter). Regional SUVR (80‐100 min and 90‐110 min) values were corrected as: SUVR
C =SUVR / [1‐β
ref/k2

,ref+β
tar
SUVR/(R
1
k2

,ref)], where SUVR
C: corrected SUVR, k2

,ref: population‐based rate constant for efflux of the radiotracer from reference region, [β
ref, β
tar]: clearance rates of the radiotracer from the reference and target tissues during SUVR time‐window, R
1=0.75 (population‐averaged value for ratio of target:reference radiotracer delivery rate). The primary outcome measures were the percent bias (%bias) in SUVR (and SUVRc) relative to DVR, and the standard deviation of %bias.

Across subjects and selected regions of interest (ROIs, Tables 1‐3), the mean %bias was 0.9% for SUVR 80‐100 min and 1.2% for SUVR 90‐110 min. The variability in %bias was 13.1% for SUVR 80‐100 min and 11.9% for SUVR 90‐110 min. The mean %bias in SUVRc across subjects and ROIs was comparable for SUVRc 80‐100 min (‐1.3%) and SUVRc 90‐110 min (‐0.1%), with similar variability (∼11%) for SUVRc 80‐100 min and SUVRc 90‐110 min. Regional measures for (averaged across subjects) are listed in Tables 1‐2. Similar performance metrics were observed for a subcohort analysis of 5 high‐binding individuals with highest average regional DVR values (1 AD,3 MCI,1 HC).

[18F]MK‐6240 SUVR in this cohort was already minimally biased; SUVRc did not significantly decrease the variability in SUVR bias. This SUVR correction method might not be beneficial for radiotracers that deviate considerably from its assumption of 1‐tissue compartment model kinetics.

Figures